Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com

August 2008 Archives

I've created a new page with Elecraft K3 audio output measurements at Elecraft K3 Receive Audio.

Last week, I purchased two LF Converter kits from Jackson Harbor Press, run by Chuck Olson, WB9KZY. I've built both kits and have added a review of the converter at Jackson Harbor Press VLF Converter. These are inexpensive kits ($14.00 plus shipping) and provide more than decent performance. I couldn't resist tinkering with the input low pass filter, however, and those readers interested in frequency selective filters may find the page interesting for that reason alone.

As part of the process, I also designed and built an AM band band-reject filter. I'll write that up in the next few days as well.
 

I'm temporarily out of stock of printed circuit boards for the Z10000 buffer amplifier. I've ordered more boards with a September 11th scheduled delivery date. Shipping will resume shortly after I receive the boards.

Dave, K4DGW, found another case where I had the wrong AGC speed mention at my http://www.cliftonlaboratories.com/elecraft_k3_agc_and_s-meter.htm page and I've fixed that this morning. At his request, I've also added several paragraphs describing the difference between fast and slow AGC in practical use.
 

I corrected the AGC fast/slow graph caption on my http://www.cliftonlaboratories.com/elecraft_k3_agc_and_s-meter.htm page. Thanks to Dave, K4DGW for catching my error.
 

I made a small change in the Elecraft K3 Noise Blanker and Crystal/DSP Filtering  page to add response curves for the 500 Hz crystal filter with the DSP set at 600 Hz.

I've added a new page Elecraft K3 Noise Blanker and Crystal/DSP Filtering presenting measurements of the K3's IF noise blanker performance (have not yet looked at the DSP noise blanker) and crystal filter and DSP filter performance, both in terms of bandwidth and also time response due to DSP processing delays. I would like to measure group delay but that's proving more difficult than I anticipated.

I'll be working on the page off and on a good part of the day, so it will be published in an incomplete stage at times  today as I test the completed portions. [page finished at 1910 EDT.]
 

I've been working on an automated method of looking at the K3's filters (both the crystal roofing filter and the DSP filter). The plot below shows the 500 Hz 5-pole crystal roofing filter and the DSP filter.

When the DSP is set to be significantly wider than the crystal filter (DSP at 4000 Hz) the filter response is dominated by the crystal filter's response. Even with the DSP set to 1500 Hz, however, the DSP response contaminates the crystal filter data once we are 40 dB or so down on the filter from the peak response.

This data was taken with only the K3 and an external oscillator; no other test equipment. I also should note that this method does not accurately show the K3's ultimate rejection. The dynamic range of this method depends on the bandwidth selected, but is 70 to 80 dB, which is more than enough to show the filter performance over the 6:60 dB "shape factor"  range.

I'll write up the procedure and make available the associated software sometime in the next week or two. I really dislike programming a Windows computers and if someone who enjoys that process more would like to work with me to develop a more refined version of my thrown together code, please drop me an E-mail message.
 

I received the following E-mail today from Doug, GM0ELP reporting his success using a Softrock Lite 6.2 as a panadapter with his Elecraft K3.

I've extensively expanded the Elecraft K3 AGC and S-Meter page to include the effects of changing the K3's AGC-F, AGC-S and AGC HOLD parameters.

I received an update from Rick, VE7TK, concerning his Ten-Tec Orion modification to extract a buffered 9 MHz IF sample for to use with an external panadapter. Rick says that he has now extended his modifications to include the Orion II (model 566) as well as the original Orion (model 565). I've modified my Z10000 Buffer Amp page accordingly. Rick's installation notes are available from his web site at http://www3.telus.net/ve7tk/9MHz_IF.pdf (contents are subject to change, as Rick notes.)

I've extensively revised the Elecraft K3 and Panadapters page to (a) reflect measurements for spurious signals into the KRX3 sub-receiver and (b) add a modification developed by Larry, N8LP, to increase the K3's IF sample output level.

I've also cleaned up a couple of sentences in the K3 AGC and S-Meter page and reformatted the two data tables.

I've written before about automatic gain control in receivers at Receiver AGC Curves but none of the receivers examined there had the flexibility of Elecraft's K3 for user-adjustable AGC slope and intercept.

I've added a page Elecraft K3 AGC and S-Meter documenting how the K3's parameters AGC THD (threshold) and AGC SLP (slope), accessible from the Configuration Menu, alter the receiver's AGC performance. I've run audio out versus RF input plots for varing AGC THD settings and AGC SLP settings.

While I had the equipment set up for these measurements, I also ran an S-meter calibration curve, comparing the K3's S-meter with RF signal levels. (Actually, "S meter" is a misnomer as the signal strength indication is part of the main LCD display.)

I've finished several days of measurements on my K3. These will lead to two new pages, one of which is now ready: Elecraft K3 and Panadapters

The new page discusses several issues related to the K3's IF sample output and how one might go about making the signal level more usable.

In addition, if you plan  to or are using a Z90 or Z91 with a K3, please read this page. Jump to the marker concerning the Z90.

I received a shipping notice from Elecraft today—my K3 is scheduled to arrive on Wednesday, the 13th. Original order date May 27, 2007, so the  total wait has been 14.5 months. The main delaying factor has been the 2nd  receiver.

The last week has  been devoted to more active antenna work, this time on the coupler end. The coupler injects 24V DC onto the coaxial cable and picks off the returned signal. Ideally,  the DC power is sufficiently decoupled from the signal so that (a) no appreciable signal loss is caused and (b) no appreciable power supply noise is injected into the signal path.

Objective (a) is not too hard to accomplish, and I have that working well. Part (b), on the other hand, is proving a bit more challenging than I thought. After the usual tricks with bypassing, chokes and the like, I still see excess noise injected from the power supply. My current assumption is that the transformer I am using has sufficient capacitive coupling between the primary and secondary windings to allow noise from the power mains to be coupled into the receiver path. I'll order a couple of split-bobbin transformers tomorrow, as that design results in much better primary/secondary isolation than the transformer I'm currently using.

On a completely different topic, I'm occasionally asked "which PIC programmer and development board should I purchase"? I've added a new page PIC Development Boards and Programmers to address these questions. This page presents my limited exposure to several boards and programmers and is not intended to be a comprehensive of the field.

The image below shows the signal level of WWVB at  60 KHz over nearly three days taken with the same conditions as previously discussed.
 

As mentioned in July updates, I've started work on a broadband (10 KHz to 30+ MHz) active antenna.

The photograph to the right shows one prototype mounted in my back yard. The whip portion is a low-band mobile radio antenna, about 50 inches (1.25 m) long.

In addition to the antenna, I've added a common mode choke with ground about 20 feet (6 m) from the antenna. The theory is that power line and computer hash noise is coupled to the braid's exterior and travels along the braid to the antenna where it is picked up by the antenna. This theory says that a choke (with ground rod) located near the antenna will reduce the noise coupling.

The choke is wound with a dozen turns of coaxial cable around a double stack ferrite core of Steward Type LF material. The choke impedance is at least several hundred ohms at 25 KHz, and rapidly increases to several thousand ohms by 500 KHz.

I've only started performance measurements, and I've noticed differences in both signal and noise amongst the possible combinations:

1. Ground at choke, none at antenna
2. Grounds at choke and antenna
3. Ground at antenna, no choke installed

My subjective impression, backed up with a quick measurement comparison is that mode (1) provides the best signal to noise ratio in the VLF and LF range. But, this conclusion may change with more data, particular when I look at higher frequencies.
 

Ron's data is limited by the 3585A's resolution, which is 1,000 data points, so over a 4 hour period, a measurement is taken once every 14.4 seconds.

My 3562A's data is taken at about 5 minute intervals, with each data point being the average of 16 readings. I've plotted both the individual 5-minute values and a smoothed average (adjacent average of 5 values on either side).

It's not possible without a lot of extra work to directly compare signal strength taken by Ron (located in suburban Detroit MI) and my data in suburban Washington DC. The relative changes are directly comparable, but not the absolute levels.

Ron provided the two images below illustrating WWVB's signal level (60 KHz) over two four-hour periods. The first is from 8 PM - midnight 01 August, whilst the second is 4 AM - 8 AM today, 02 August 2008. (All times are EDT).

Studies of VLF/LF propagation have long showed anomalies at sunrise and sunset. (Local sunset at Ron's is 8:50 PM EDT this time of year.) The studies usually show a dip in signal level, followed by an increase at local (receiver) sunset. In some cases, a second dip is seen at local sunset at the transmitter location, in this case in Colorado, near Ft. Collins. Local sunset at WWVB is 8:13 MDT, or 10:13 EDT. At sunrise, the reverse effect is seen.

In the 8 PM - 12 midnight plot below, local (receiver site) sunset is approximately 2.5 divisions from the left edge, and increased signal levels are seen shortly after sunset. Transmitter sunset is essentially mid-screen, at which point the signal level flattens and exhibits a dip of perhaps 1 to 1.5 dB.

Receiving sunrise corresponds to 6 divisions from the left, and we indeed see a major drop in signal around local sunrise, with a 10 dB decrease in signal level over 30 minutes or so. Sunrise at the transmitter occurs at the end of the data, so we can not say too much about it.

It's also apparent that Ron's local noise dropped considerably between midnight and 0400—the "fuzz" on the data plot is noticeably reduced in the plot below.

Ron's data shows a 10 dB increase in signal level over the space of about 2.5 hours. I see a similar signal increase over about a 2 to 2.5 hour period. Our sunrise observations are similar as well.

I observed three peaks during the hours between sunset and sunrise, with about 5 db peak/valley changes. Ron's data also shows peak/valley changes of about 4 db, with perhaps a slightly shorter peak-to-peak time.

The strongest VLF station on the east coast is NAA's 1 megawatt ERP signal at 24 KHz. Over the same period, it shows faster increases/decreases at sunrise/sunset. Of course, sunrise/sunset at NAA's Maine transmitter is within 30 minutes of sunrise/sunset in Clifton VA, so a more rapid rate of change in signal level makes sense.

This path also shows similar characteristics, with a slow increase at sunset and a rather fast drop at sunrise.